Used to do military RF and industrial radio control/comms for a few years so I'll add in some Fun Facts. Apologies for the long diatribe.
- Cross-polarising dipoles (ie having one vertical and the other 90 degrees horizontal but still perpendicular to the other) will nominally drop your RF levels by 30dB. Useful if/when your RF levels are too high (also works for Yagi antennas as well but here's other tricks 'n traps with that style of stick that can be exploited, but not here...).
- Having a dipole point directly up along it's length at your destination/originating signal source will drop it metric shitloads, as doing that you'll end up trying to use the worse possible portion of the antenna's doughnut-shaped propagation pattern - you'll actually be relying on reflected signal pathing from objects/structures within the RF pathing fields. Useful if the levels are stoopidly too high. Also useful to help directionalise the RF energy when using dipoles.
- Dipoles work exactly the same pointing straight up or straight down. The only differences are in the actual RF signal pathing and placement of any obstacles and nearby RF reflective/absorbitive surfaces etc when you place the dipole in what 99.999% of the public consider to be such an unnatural orientation. For example - if you want to fit a stick on your house, then putting aside the drop in vertical height (and potential impact on Line Of Sight, which all antennas love) and the wall then behind it, there's nothing wrong installing a dipole up under the eaves of the roof pointing downwards.
- RF can be one of the most insidious forms of energy known to Man, ask anyone who's been involved with shield rooms for example. If you've got issues there with sensitivity to RF then look at repositioning the antennas (thence altering propagation patterning/RF pathing) such as the help minimise any effect you feel - for example put an antenna under the desk, or up on the top of the cupboard well above your head height. Even up into the roof cavity.
- Oh, and be aware home-rolling coax leads can be problematic, as it's possible to cut coax lengths such as to act as ideal RF paths, dead shorts or even open circuits, all based on actual conductor lengths vs RF frequency.
Back to one of your original problems.
- If you want to drop RF transmissions levels even further than the minimum Tx setting in your router, have a go at adding a suitable RF attenuator onto the coax lead. But be aware this will affect RF levels in both directions, Tx and Rx.
- A reeeeally sneaky trick along these lines is to not screw the connector all the way home, leaving a small but finite air gap 'tween the coax and antenna centre pins. Be aware though, apart from being somewhat Trial and Error With Associated Fiddling this can also become somewhat icky for consistent/long-term stability if you're reeeeeally anal about the numbers - temperature variations in the metals involved for example can make quite significant variations in this air-gap attenuation over time.
- If you simply (!) want to drop Tx levels but keeps the Rx sensitivity levels up then you can start looking at either a router/WiFi device with separate fixed or configurable Tx and Rx ports, then putting a small or lower-gain antenna on the Tx and a higher gain antenna on the Rx one.
- Or start to look at fiddling with Diplexers. and different gains of antennas for the Tx and Rx ports.
Now noting an important item in your last post - "...often see 3 of 5bars of signal, but get no reply..."
Rule 1 - when connectors are involved, always check that the centre pin isn't bent or been pushed back, and always check for kinks or sharp bends in any coax as well. Either of these will screw up the RF levels/sensitivities, only needs one at the right end to give you grief.
Rule 2 - check the RF numbers/levels at BOTH ends of the radio paths and compare each end. Ideally they should be equal for both devices.
Rule 3 - ideally swap the two devices end-for-end on the troublesome radio path as well, then recheck the numbers - they should follow the devices around. If they don't then you might have external RF interference affecting Rx sensitivity at one end of the path.
Rule 4 - do not assume the devices (both routers/wifi cards and even the antennas themselves) have the same Rx sensitivities, nor that the configured Tx levels/numbers and listed Gain numbers (on antennas) are actually the truth, until proven otherwise. Trust Me On This - I've encountered oodles of both Yum Cha and Name/Brand WiFi antennas and various radio equipment at various frequencies (from near DC to microwave) that under real testing turn out to be basically stuffed/misleading...
If you're experiencing acceptable Rx levels at one end but not the other then a simple fix can sometimes be add a higher gain antenna at the lower Rx level end.
Anything not nailed down is mine, and anything I can prise up is not nailed down...